into zippy.cornsilk.net:/home/cmiller/work/mysql/mysql-4.1-maint
configure.in:
Auto merged
mysql-test/t/ps.test:
Auto merged
sql/handler.cc:
Auto merged
sql/sql_delete.cc:
Auto merged
sql/sql_select.cc:
Auto merged
sql/table.cc:
Auto merged
tests/mysql_client_test.c:
Auto merged
myisam/sort.c:
Manual merge.
mysql-test/r/innodb_mysql.result:
Manual merge.
mysql-test/t/innodb_mysql.test:
Manual merge.
mysys/mf_iocache.c:
Manual merge.
- Because my_seek actually is capable of returning an error code we should
exploit that in the best possible way.
- There might be kernel errors or other errors we can't predict and capturing
the return value of all system calls gives us better understanding of
possible errors.
mysys/mf_iocache.c:
- Added check on return value for my_seek
- Added comments
mysys/my_chsize.c:
- Added check on return value for my_seek
- Added comments
mysys/my_lock.c:
- Added check on return value for my_seek
- Added comments
mysys/my_seek.c:
- Added comments
into chilla.local:/home/mydev/mysql-4.1-bug8283-one
myisam/mi_check.c:
Auto merged
myisam/mi_packrec.c:
Auto merged
myisam/sort.c:
Auto merged
mysql-test/r/myisam.result:
Bug#8283 - OPTIMIZE TABLE causes data loss
Manual merge
mysql-test/t/myisam.test:
Bug#8283 - OPTIMIZE TABLE causes data loss
Manual merge
OPTIMIZE TABLE with myisam_repair_threads > 1 performs a non-quick
parallel repair. This means that it does not only rebuild all
indexes, but also the data file.
Non-quick parallel repair works so that there is one thread per
index. The first of the threads rebuilds also the new data file.
The problem was that all threads shared the read io cache on the
old data file. If there were holes (deleted records) in the table,
the first thread skipped them, writing only contiguous, non-deleted
records to the new data file. Then it built the new index so that
its entries pointed to the correct record positions. But the other
threads didn't know the new record positions, but put the positions
from the old data file into the index.
The new design is so that there is a shared io cache which is filled
by the first thread (the data file writer) with the new contiguous
records and read by the other threads. Now they know the new record
positions.
Another problem was that for the parallel repair of compressed
tables a common bit_buff and rec_buff was used. I changed it so
that thread specific buffers are used for parallel repair.
A similar problem existed for checksum calculation. I made this
multi-thread safe too.
include/my_sys.h:
Bug#8283 - OPTIMIZE TABLE causes data loss
Redesign of io_cache_share.
include/myisam.h:
Bug#8283 - OPTIMIZE TABLE causes data loss
Redesign of checksum calculation in mi_check.c.
'calc_checksum' is now in myisamdef.h:st_mi_sort_param.
myisam/mi_check.c:
Bug#8283 - OPTIMIZE TABLE causes data loss
Implemented a new parallel repair design.
Using a synchronized shared read/write cache.
Allowed for thread specific bit_buff, rec_buff, and calc_checksum.
myisam/mi_open.c:
Bug#8283 - OPTIMIZE TABLE causes data loss
Added DBUG output.
myisam/mi_packrec.c:
Bug#8283 - OPTIMIZE TABLE causes data loss
Allowed for thread specific bit_buff and rec_buff.
myisam/myisamdef.h:
Bug#8283 - OPTIMIZE TABLE causes data loss
Commented on checksum calculation variables.
Allowed for thread specific bit_buff.
Added DBUG output for better table crash detection.
myisam/sort.c:
Bug#8283 - OPTIMIZE TABLE causes data loss
Added implications of the new parallel repair design.
Renamed 'info' -> 'sort_param'.
Added DBUG output.
mysql-test/r/myisam.result:
Bug#8283 - OPTIMIZE TABLE causes data loss
Added test results.
mysql-test/t/myisam.test:
Bug#8283 - OPTIMIZE TABLE causes data loss
Added test cases.
mysys/mf_iocache.c:
Bug#8283 - OPTIMIZE TABLE causes data loss
Redesign of io_cache_share.
We do now allow a writer to synchronize himself with the
readers of a shared cache. When all threads join in the lock,
the writer copies the data from his write buffer to the shared
read buffer.
into mysql.com:/home/svoj/devel/mysql/merge/mysql-4.1-engines
mysql-test/r/myisam.result:
Auto merged
mysql-test/t/myisam.test:
Auto merged
sql/table.cc:
Auto merged
Crash may happen when selecting from a merge table that has underlying
tables with less indexes than in a merge table itself.
If number of keys in merge table is not bigger than requested key number,
return error.
myisammrg/myrg_open.c:
Store min(number of keys) in m_info instead of number of keys in last
underlying table.
myisammrg/myrg_queue.c:
Return error if inx passed to _myrg_init_queue function is not less
than number of keys.
mysql-test/r/merge.result:
A test case for bug#21617.
mysql-test/t/merge.test:
A test case for bug#21617.
mysys/queues.c:
Replaced annoying ifndef DBUG_OFF with DBUG_ASSERT, fixed coding style.
The problem was that having queue overrun in debug build was hidden
with this ifdef.
into siva.hindu.god:/usr/home/tim/m/bk/tmp/41
mysys/my_lwrite.c:
Auto merged
mysys/my_pread.c:
Auto merged
mysys/my_read.c:
Auto merged
mysys/my_write.c:
Auto merged
- Fix my_read/my_write to handle return values from read/write correctly
- Add debugging 'deprecated function' warning to my_lread/my_lwrite
- Add debugging 'error, read/write interrupt not handled' warning to my_quick_read/my_quick_write
There is no test case associated with these changes. However, this is a conservative change,
and no repeatable test case is available.
mysys/my_lread.c:
Warn about using deprecated function.
mysys/my_lwrite.c:
Warn about using deprecated function.
mysys/my_pread.c:
Handle interrupted read() or write() (EINTR) properly
mysys/my_quick.c:
Warn about interrupted read() or write(), which is not
handled by my_quick_read() or my_quick_write().
mysys/my_read.c:
Handle interrupted read() (EINTR) properly
mysys/my_write.c:
Handle interrupted write() (EINTR) properly
The problem is that on some Mac OS X-es the file writing/reading
call with zero bytes to read/write returns error.
So here i try to eliminate that kinds of calls.
myisam/mi_check.c:
zero length copying avoided
mysys/my_chsize.c:
no file operations if it's not necessary
event' from master"
Since there is no repeatable test case, and this is obviously wrong, this is
the most conservative change that might possibly work.
The syscall read() wasn't checked for a negative return value for an
interrupted read. The kernel sys_read() returns -EINTR, and the "library"
layer maps that to return value of -1 and sets errno to EINTR. It's
impossible (on Linux) for read() to set errno EINTR without the return
value being -1 .
So, if we're checking for EINTR behavior, we should not require that the
return value be zero.
mysys/my_read.c:
The read() syscall should check for negative one, since that (usually) signals
errors (like being interrupted) and zero (usually) signals end-of-file .
Too many cursors (more than 1024) could lead to memory corruption.
This affects both, stored routines and C API cursors, and the
threshold is per-server, not per-connection. Similarly, the
corruption could happen when the server was under heavy load
(executing more than 1024 simultaneous complex queries), and this is
the reason why this bug is fixed in 4.1, which doesn't support
cursors.
The corruption was caused by a bug in the temporary tables code, when
an attempt to create a table could lead to a write beyond allocated
space. Note, that only internal tables were affected (the tables
created internally by the server to resolve the query), not tables
created with CREATE TEMPORARY TABLE. Another pre-condition for the
bug is TRUE value of --temp-pool startup option, which, however, is a
default.
The cause of a bug was that random memory was overwritten in
bitmap_set_next() due to out-of-bound memory access.
mysys/my_bitmap.c:
Local 'bitmap_size' is measured in bytes, no need to multiply it by 8.
sql/sql_select.cc:
Clear the temp_pool_slot bit only if we have set it previously.
hp_test2 now works again
Fixed wrong cast, which caused problems with gcc 4.0 and floats in prepared statements (Bug #19694)
heap/hp_test1.c:
Portability fix
heap/hp_test2.c:
Added max_table_size (fixes that hp_test2 works again)
include/my_global.h:
Fixed wrong cast, which caused problems with gcc 4.0 (Bug #19694)
mysys/my_handler.c:
Added missing include file
strings/strtod.c:
Fixed include files
- Check that length of value is longer than 1 before decrementing length by 2.
- Backport from 5.0, make it possible to use my_print_defaults in tests
mysql-test/mysql-test-run.pl:
Backport from 5.0, make it possible to use my_print_defaults from tests
mysql-test/mysql-test-run.sh:
Backport from 5.0, make it possible to use my_print_defaults from tests
mysql-test/r/mysqldump.result:
Update result
mysql-test/t/mysqldump.test:
Test that my_print default don't segfault when encountering an option without closing "
mysys/default.c:
Check that length of value is longer than 1 before deciding to decrement its length by 2.
mysql-test/std_data/bug15328.cnf:
New BitKeeper file ``mysql-test/std_data/bug15328.cnf''
The bug caused a reported index corruption in the cases when
key_cache_block_size was not a multiple of myisam_block_size,
e.g. when key_cache_block_size=1536 while myisam_block_size=1024.
mysql-test/r/key_cache.result:
Added a test case for bug #19079.
mysql-test/t/key_cache.test:
Added a test case for bug #19079.
A wrong cast led to numeric overflow for data files
greater than 4GB. The parallel repair assumed end of
file after reading the amount of data that the file
was bigger than 4GB. It truncated the data file and
noted the number of records it found so far in the
index file header as the number of rows in the table.
Removing the cast fixed the problem.
I added some cosmetic changes too.
The normal repair worked because it uses a different
function to read from the data file.
mysys/mf_iocache.c:
Bug#11527 - Setting myisam_repair_threads to >1 leads to corruption
The pure fix was to remove a cast from a file offset difference.
Supplemented this with warnings in function comments,
a change from == to <= to be slightly more safe,
a renaming from "read_len" to "left_length" to make the
partial code duplication between _my_b_read() and _my_b_read_r()
more obvious and easier to compare the functions,
removed another unnecessary (but harmless) cast,
and fixed coding sytle around the "left_length" changes.
rename sha1_* to mysql_sha1_*
include/sha1.h:
rename sha1_* to mysql_sha1_*
mysys/sha1.c:
rename sha1_* to mysql_sha1_*
sql/item_strfunc.cc:
rename sha1_* to mysql_sha1_*
sql/password.c:
rename sha1_* to mysql_sha1_*
when high concurrency": remove HASH::current_record and make it
an external search parameter, so that it can not be the cause of a
race condition under high concurrent load.
The bug was in a race condition in table_hash_search,
when column_priv_hash.current_record was overwritten simultaneously
by multiple threads, causing the search for a suitable grant record
to fail.
No test case as the bug is repeatable only under concurrent load.
include/hash.h:
- remove current_record from HASH, instead modify hash_first,
hash_next to accept HASH_SEARCH_STATE as an IN/OUT parameter
mysys/hash.c:
- remove HASH::current_record
- change declarations of functions that use HASH in read-only mode
to accept const HASH * instead of HASH *.
- implement hash_search; move the old implementation of hash_search
to hash_first
mysys/testhash.c:
- adjust the test case to changed function declarations
sql/lock.cc:
- adjust to changed declarations of hash_search, hash_next
sql/sql_acl.cc:
- adjust to changed declarations of hash_search, hash_next
sql/sql_base.cc:
- adjust to changed declarations of hash_search, hash_nex
sql/sql_cache.cc:
- adjust to a changed declaration of hash_replace
several times, it was causing corruption of charset data stored in once_mem_pool.
my_end() deallocated once_mem pool, but did not put a flag that charsets have
to be reloaded. The fix addresses this problem.
options to the wrong value. (Bug #12925)
mysql-test/t/mysql_client_test.test:
Add parameter for testing getopt bug
mysys/my_getopt.c:
Remove incorrect and unnecessary casts
tests/mysql_client_test.c:
Add test case for Bug #12925 (my_getopt bug)
heap/_check.c:
Change arguments to ha_key_cmp
heap/hp_create.c:
Change arguments to ha_key_cmp
include/my_base.h:
Remove SEARCH_RETURN_B_POS and instead always send an array to ha_key_cmp() as last argument
myisam/mi_check.c:
Change arguments to ha_key_cmp
myisam/mi_rnext_same.c:
Change arguments to ha_key_cmp
myisam/mi_search.c:
Change arguments to ha_key_cmp
myisam/mi_write.c:
Change arguments to ha_key_cmp
myisammrg/myrg_queue.c:
Change arguments to ha_key_cmp
mysys/my_handler.c:
Remove SEARCH_RETURN_B_POS and instead always send an array to ha_key_cmp() as last argument
(This removes an if in a loop at the expensive of an int on the stack)
sql/records.cc:
Simplify new rr_index() code
Create common error handling function for rr_() functions.
Remove loop from rr_index() as handler::index_next() can never return HA_ERR_RECORD_DELETED
sql/sql_load.cc:
Simplify
sql/sql_update.cc:
Simplify code
Fixed bug when one is updating an index column that could be used with ORDER BY
sql/structs.h:
Removed not needed structure element
Bug#13238 mysqldump and mysqladmin hangs
Avoid recursion into init_available_charsets.
Serg's version didn't work for me: I got
double mutex locking.
Pushing this version instead
(It was approved by Serg anyway)
mysys/charset.c:
Bug#13238 mysqldump and mysqladmin hangs
Avoid recursion into init_available_charsets.
Serg's version didn't work for me: I got
double mutex locking.
Pushing this version instead
(It was approved by Serg anyway)
during replication.
Modified my_dir(). Now this function skips hidden and system
files which sometimes are created by Windows.
NOTE. The fix is similar to the previuos one (05 July 2004)
except for correct setting of the 'attrib' variable value
(within the previous fix this variable was left uninitialized
when my_dir() was called with My_flags & MY_WANT_STAT == 0).
mysys/my_lib.c:
Modified my_dir().
Now this function skips hidden and system files
which sometimes are created by Windows.
signal handlers are set up, the blocking flags for sockets are set,
and which thread-related functions are used. (Bug #8731)
configure.in:
Fix flags for Darwin 6 and later. Simplify Darwin 7-9 blocks to simply
be a catch-all for *darwin* so that future Darwin releases get the
latest flags.
include/config-win.h:
Define my_sigset() instead of sigset().
include/my_pthread.h:
Define my_sigset() instead of trying to monkey with sigset(), and favor
an implementation based on sigaction().
mysys/my_pthread.c:
Remove pthread_signal(), which is identical to the new my_sigset() macro.
mysys/thr_alarm.c:
Use my_sigset() instead of sigset().
sql/mysqld.cc:
Use my_sigset() instead of signal() and sigset(), remove unnecessary
definition of sigset on __amiga__. Remove unused THREAD_SPECIFIC_SIGPIPE
code.
A future improvement would be to re-assess the use of sigaction() here
and convert its usage to use my_sigset().
vio/vio.c:
Always call fcntl() to initialize flags of socket in initialization to
avoid problems on systems that don't report the flags on a socket
correctly right after it has been returned from accept(), such as
FreeBSD, Mac OS X, and possibly other BSD-derived systems.
vio/viosocket.c:
If fcntl() fails in vio_blocking(), restore the flags stored in the
vio struct.
mysql-test/r/wait_timeout.result:
New BitKeeper file ``mysql-test/r/wait_timeout.result''
mysql-test/t/wait_timeout-master.opt:
New BitKeeper file ``mysql-test/t/wait_timeout-master.opt''
mysql-test/t/wait_timeout.test:
New BitKeeper file ``mysql-test/t/wait_timeout.test''
mysys/my_init.c:
Changed implementation slightly to use a local variable for keeping track of wheter we should print to the info_file even if the flag is not set in infoflag
Explicitly turn off print_info if there is no DBUG_FILE
a restriction to 255 chars for editable buffer.
VC++Files/mysys/mysys.dsp:
Added my_conio.c
VC++Files/mysys/mysys_ia64.dsp:
Added my_conio.c
include/my_sys.h:
Added declarations for my_conio.c functions
mysys/my_conio.c:
Added _cgets() replacement that is not limited to 255 chars retrieval
from win32 console.
After merge fix.
include/keycache.h:
Bug#12920 - key_read_requests counter appears to re-set
After merge fix.
Status variables moved into KEY_CACHE in 4.1.
Enlarging them there.
include/my_sys.h:
Bug#12920 - key_read_requests counter appears to re-set
After merge fix.
Status variables moved into KEY_CACHE in 4.1.
Removed unused global variables.
myisam/mi_test2.c:
Bug#12920 - key_read_requests counter appears to re-set
After merge fix.
Status variables moved into KEY_CACHE in 4.1.
Changed to print default key cache status.
mysys/mf_keycache.c:
Bug#12920 - key_read_requests counter appears to re-set
After merge fix.
Status variables moved into KEY_CACHE in 4.1.
Changed their debug printout.
sql/mysqld.cc:
Bug#12920 - key_read_requests counter appears to re-set
After merge fix.
Changed to print full value.
sql/sql_show.cc:
Bug#12920 - key_read_requests counter appears to re-set
After merge fix.
Added a branch for long long values.
sql/sql_test.cc:
Bug#12920 - key_read_requests counter appears to re-set
After merge fix.
changed to print full values.
sql/structs.h:
Bug#12920 - key_read_requests counter appears to re-set
After merge fix.
Added a branch for long long keycache values.
on Windows. (Bug #12325)
mysql-test/r/lowercase_table.result:
Update results
mysql-test/t/lowercase_table.test:
Move test to new windows.test, since it is Windows-specific
mysys/my_access.c:
Check that we've matched the whole forbidden name.
mysql-test/include/windows.inc:
New BitKeeper file ``mysql-test/include/windows.inc''
mysql-test/r/windows.result:
New BitKeeper file ``mysql-test/r/windows.result''
mysql-test/t/windows.test:
New BitKeeper file ``mysql-test/t/windows.test''
